- Neither layout wins on its own. Baymard's usability testing found shoppers move between steps without much trouble. The failures come from what each step asks for, not from the number of steps.
- Field count is the lever the research supports. A good checkout works with 12-14 form elements. The average US checkout shows 23.48. You can hit 23 on one page as easily as on four.
- The credible one-page lift is small. Shopify reports roughly 0.5% on average, one merchant case study reports 3.5%, and a two-week agency comparison reports 7.5%. The "66% fewer abandoned carts" figures come from vendors selling the thing being measured.
- One-page tends to win on simple, low-value, mobile-heavy, wallet-heavy orders where the total can be shown early.
- Multi-step tends to win when later fields depend on earlier answers, orders are complex or B2B, or you need the email before the shipping cost appears.
- Four trade-offs almost nobody prices in: abandoned-cart email capture, funnel granularity in analytics, checkout page weight on mid-range Android, and where your upsells can go.
Neither layout wins on its own. The number of pages in your checkout is not the variable that decides whether people complete the purchase. What you ask them to do is. A well-built three-step checkout will beat a cluttered one-page checkout, and a well-built one-page checkout will beat a bloated three-step one. The research this debate is usually built on says exactly that, and has said it for over a decade.
That said, the layouts are not interchangeable. There are identifiable conditions under which each one tends to perform better, and there are real trade-offs in email capture, funnel measurement and page performance that most comparisons never mention. This article covers what the evidence actually supports, where the widely quoted numbers come from, and how to make the call for your specific store.
What counts as one-page, multi-step and accordion?
Three patterns are in common use, and a lot of confusion comes from people using the same word for different things.
Multi-step (multi-page) checkout splits the flow across separate pages with separate URLs: contact details, then shipping, then payment, then an order review. Each step submits and loads the next.
One-page checkout collects everything needed to complete the order on a single scrollable page: contact, shipping address, shipping method, payment and the order summary.
Accordion checkout sits between the two. All sections live on one page, but only one is open at a time, and completed sections collapse as the shopper moves down. Visually it reads as one page. Behaviourally, shoppers treat it as a multi-step flow.
That third pattern matters more than the binary suggests, because it is what several major platforms actually ship. Magento 2's default checkout is frequently described as one-page, but it presents two sequenced steps (shipping, then review and payment) on a single URL, which makes it an accordion in practice. Shopify's current default is a single page where completed sections roll up into one-line summaries. So for a large share of stores the real question is not "should we go one-page?" It is "our platform already ships something one-page-ish; is it configured well?"
What does the research actually say?
The most-cited source in this debate is Baymard Institute, whose checkout usability research underpins most serious ecommerce UX work. Their position on one-page checkout is unusually blunt and rarely quoted accurately.
Baymard's research director, Christian Holst, addressed it directly. A/B tests do exist in which one-page checkouts substantially outperform multi-step ones, but those tests typically pit an un-optimised multi-step checkout against a newly optimised one-page checkout. In Baymard's own usability testing, participants had relatively few problems moving between multiple steps. The usability failures came from what shoppers were asked to do at each step, not from the existence of the steps. His conclusion: take the exact same set of form fields from a one-page checkout, split them across two pages, and you would expect little or no difference in completion rate.
The rest of Baymard's checkout data points the same direction. Their survey of abandonment reasons, excluding shoppers who were simply browsing, puts extra costs such as shipping, tax and fees at the top at 40%, followed by slow delivery at 20%, distrust of the site with card details at 19%, and a forced account requirement at 18%. "Too long or complicated checkout process" sits at 17%, tied with site errors and crashes.
That 17% is the figure the one-page camp usually reaches for. But it is worth reading what Baymard attaches it to: their testing indicates an ideal checkout flow can be as short as 12-14 form elements (7-8 counting only the fields), while the average US checkout displays 23.48 elements by default (14.88 fields). "Too long" is a count of fields, not a count of pages. You can hit 23 form elements on one page just as easily as on four. You just have to scroll further to see them.
The single most quoted checkout statistic in the industry, that the average large ecommerce site can gain a 35.26% conversion increase through better checkout design, comes from this same body of work, and it refers to fixing documented usability issues across the whole flow. It is not a one-page-versus-multi-step result, though it is routinely presented as one.
Where do the "one-page lifts conversions by X%" numbers come from?
If you search this topic, you will find confident percentages ranging from 0.5% to over 60%. They cannot all be true. Tracing them back is instructive, because the pattern is consistent: the further a number travels from its source, the bigger it gets.
| Claim | Actual source | What it really measures |
|---|---|---|
| ~0.5% conversion lift | Shopify's own reported average for merchants moving to one-page checkout | Platform-wide average across a very large, very mixed merchant base |
| 3.5% conversion lift | Shopify case study, Stellar Eats (a Canadian food brand) | One merchant, self-reported, in a vendor-published case study |
| 7.5% conversion lift | Digismoothie, 2023 | One store, roughly two weeks of data, period-over-period rather than a controlled A/B test, as the authors themselves state openly |
| "20-30%", or "up to 66% fewer abandoned carts" | Checkout extension and app vendors | Marketing copy, no published methodology, usually selling the thing being measured |
| "21% increase, per Baymard" | Various SEO blogs | Not found in any Baymard primary source we could locate. Baymard's published position is the opposite |
The honest reading of the credible end of that range: switching to a well-implemented one-page checkout is worth something on the order of a fraction of a percent to a few percent of conversion rate for a typical store. That is real money at scale, and worth doing, but not the transformation the louder numbers imply.
There is also evidence in the other direction that rarely gets cited. When Shopify made one-page checkout the default, a substantial thread of merchants reported conversion declines after the switch, some severe. Shopify's product team responded that the layout had been tested globally for more than a year and improves conversion on average, and that they had found no systemic errors. Both things can be true: an average lift across hundreds of thousands of stores is entirely compatible with meaningful losses at some of them. "On average" is doing a lot of work in this debate, and your store is not an average.
When does one-page checkout tend to win?
Based on the mechanism rather than the marketing, one-page tends to perform better when:
- The order is simple. A few line items, one shipping address, one delivery option, no configuration. Nothing needs sequencing because nothing depends on anything else.
- Traffic is mobile-dominant and impulse-led. Fewer page loads means fewer opportunities for a flaky connection to drop someone mid-flow.
- Average order value is low enough that the purchase is not deliberated. Nobody wants a considered, reassuring, multi-stage experience when buying a $28 item.
- Express wallets carry a large share of checkouts. If most shoppers complete via Apple Pay, Google Pay or Shop Pay, the form layout underneath matters less, and one page keeps the manual path short for everyone else.
- Your shipping cost can be calculated and shown early. Much of one-page's real advantage is that the total becomes visible sooner, which attacks the actual top abandonment cause rather than the page count.
When does multi-step checkout tend to win?
Multi-step, or an accordion with clear stages, tends to hold up better when:
- Later fields depend on earlier answers. Shipping method options that change by address, tax or duties that resolve by destination, delivery-date pickers tied to a postcode. Forcing these onto one page produces fields that appear, shift and re-render while the shopper is typing.
- The order is complex or high-value. Multiple addresses, gift options, scheduled delivery, installation, configurable products. Splitting related decisions is a legitimate way to reduce simultaneous cognitive load. Not because shoppers cannot handle steps, but because they cannot handle making four unrelated decisions at once.
- B2B or regulated flows require extra data. Purchase-order numbers, VAT and tax-exemption details, approval workflows, account terms. This information genuinely does not belong next to a card form.
- You need email captured before the shopper sees the shipping cost. More on this below. It is the most underrated trade-off in the entire discussion.
- Your checkout is heavy. If the page must load several payment SDKs, a tax service, an address validator and a carrier-rates call, doing all of it at once on one page can be slower than staging it.
Two of those conditions are worth grounding in work rather than argument. For Stran, a B2B promotional products business, the checkout had to carry quote workflows, approval routing and per-item personalisation fields. The decision complexity was the product, and a single form could not hold it. For Farmasi, operating across multiple markets, the staged flow was what made it possible to surface the right trust signals, currency and language context at the point each one became relevant, instead of stacking every market's reassurance onto one screen. That second point generalises: steps give you somewhere to put trust signals where they are contextually earned, which a single screen makes you fight for.
Side-by-side comparison
| Factor | One-page | Multi-step or accordion |
|---|---|---|
| Best fit | Simple, low-AOV, mobile-heavy, impulse purchases | Complex, high-AOV, B2B, conditional logic |
| Perceived effort | Lower: the end is visible | Higher per page, lower per screen |
| Cost transparency | Total visible sooner | Often revealed later, which is a real risk |
| Email capture for recovery flows | Weaker: shoppers can see shipping costs and leave before entering contact details | Stronger: contact info is typically captured at step one |
| Funnel measurement | Coarser: fewer natural step boundaries to instrument | Finer: each step is a measurable drop-off point |
| Page performance | All scripts and SDKs load at once | Load can be staged across steps |
| Back-button behaviour | Frequently broken unless the History API is implemented | Native and predictable |
| Error handling | One failure can disturb the whole page | Failures contained within a step |
| Implementation risk | Higher on platforms where it means replacing native checkout | Lower: usually the platform default |
The accordion middle ground, and the one mistake that ruins it
For most stores the practical answer is an accordion: the compactness of one page, with the sequencing of multiple steps. Baymard's testing supports this pattern, with one specific condition.
Completed sections must collapse into summaries, not just headers. If a finished section collapses to the words "Shipping Address" with no visible content, shoppers who want to double-check their details (and Baymard's eye-tracking shows they routinely do) must reopen each section one at a time. Accordion checkouts usually have no separate order-review step, so those collapsed summaries are the only review mechanism the shopper has. Strip the content out of them and you have removed the review step without replacing it.
The second requirement is the browser Back button. Baymard's testing found that if a view looks or feels like a new page, shoppers treat it as one, and they reach for Back to return to a previous step regardless of how it is implemented underneath. On a single-URL accordion, that sends them all the way back to the cart. The fix is history.pushState(): change the URL as sections advance, without a page reload, so Back does what shoppers expect.
The hybrid pattern: one-page on desktop, multi-step on mobile
The whole debate assumes one answer has to serve every visitor, and it does not. Desktop and mobile are different constraints wearing the same URL. A desktop viewport can hold a full checkout form without turning it into a tunnel. A 390px phone cannot, and the same form that reads as compact on a laptop becomes a long scroll where shoppers lose their place and pass required fields without noticing.
So render both. One page on desktop, where the space exists, and a stepped flow with a visible progress indicator on mobile, where each step fits the viewport. We built exactly this in our ecommerce checkout transformation: a single-page desktop layout, a multi-step mobile flow with a progress bar, and PayPal, Klarna and Afterpay wired across both paths.
Two caveats before you reach for it. First, you are now maintaining and instrumenting two flows, which doubles the surface area for the funnel-measurement problem described below, and your analytics has to segment by device or the aggregate will tell you nothing. Second, it only pays off if your device split is genuinely mixed. If the large majority of your checkouts are mobile, build the mobile flow properly and serve it everywhere: two flows is a real maintenance cost, and it should buy you something.
The trade-offs nobody mentions
Four consequences of this decision fall outside UX entirely, and they are the ones that tend to surprise teams after the change ships.
Abandoned-cart email capture. A multi-step checkout usually collects an email address at step one, before shipping costs are shown. A one-page checkout can display shipping costs while the contact field is still empty. Shoppers who balk at the shipping cost now leave without you ever having their address, so they never enter the recovery flow. This is a documented complaint from merchants after Shopify's default change. If recovery email drives a meaningful share of your revenue, model that loss before you move, because it can quietly offset the conversion gain.
Funnel granularity. Multi-step checkouts give you natural instrumentation points. Collapse to one page and add_shipping_info and add_payment_info either fire in quick succession or need custom implementation to fire at all, so "shoppers drop at the shipping step" becomes much harder to see. Definitions also shift: what counts as "reached checkout" may not be the same event before and after the change, which quietly invalidates your period-over-period comparison. This is precisely why so many of the public numbers in this debate are unreliable.
Performance. One page means every payment SDK, address validator, tax service and carrier-rate call initialises together. On a fast connection nobody notices. On mid-range Android over 4G, which is a large share of real traffic, this is where a "faster" checkout becomes slower. Measure checkout LCP separately from the rest of the site. It is usually the slowest page you own, and the one nobody tests.
Where the upsells go. A multi-step flow has natural insertion points between stages: a recommendation after the cart summary, a gift-wrap option on the shipping step, an extended warranty on the order review. Collapse to one page and those slots disappear. Anything you add now competes for attention with the form itself, which is the one thing you need the shopper to finish, so the offer usually gets pushed back to the cart page or forward to the order confirmation. Neither placement is wrong, but if a meaningful share of your average order value comes from in-checkout attach, price that in before you flatten the flow.
How should you decide for your store?
The decision is empirical, and most teams skip the step that makes it answerable.
- Instrument the flow before you change it. If you cannot currently see where shoppers drop out, you will not be able to tell what the change did. Verify your checkout events fire once each, reconcile your analytics purchase count against your payment processor, and check that consent settings are not silently removing EU traffic from the funnel. The same checks open our diagnostic for a post-redesign conversion drop, for the same reason.
- Count your form elements, not your pages. If your default flow shows more than about 14 elements, that is the higher-value fix regardless of layout, and it is the one the research most clearly supports.
- Check the actual abandonment reasons first. Shipping cost, delivery speed, trust and forced account creation account for far more lost orders than page count. Changing layout while leaving a forced-account-creation wall in place is rearranging furniture.
- Then test, and segment the result. Split by device and by customer type at minimum. A one-page layout that helps mobile guests can hurt returning desktop buyers with saved addresses, and an aggregate result will show you neither. Before reading the result, know what a normal completion rate looks like for your kind of store.
- Be honest about what your test can support. Two weeks of period-over-period data on one store, during a promotional period, is not evidence. If you cannot run a controlled test, say so when you report the result. Most of the noise in this topic exists because somebody did not.
What about Magento 2 and Shopify specifically?
On Shopify, one-page is the default and, for non-Plus plans, has generally not been switchable. That makes the layout debate mostly academic. The work is in configuring what is there: field reduction, express wallets, early shipping-cost display, and handling the email-capture gap in your recovery flows. Plus merchants have more control through checkout extensibility. Our Shopify checkout optimisation service covers what each plan can and cannot change.
On Magento 2, the default is already an accordion on a single URL, and the "one-step checkout" extension market exists to flatten it further. Proceed carefully here. These extensions replace a core, heavily customised part of the platform, and the recurring failure modes are well documented: conflicts with other checkout extensions, payment and shipping integrations breaking, duplicated or vanished fields, and slower checkout page loads. The vendor claims attached to them ("66% fewer abandoned carts") should be treated as marketing, not measurement. If your Magento checkout is underperforming, the field count, the shipping-rate call and the caching configuration are almost always the larger and safer wins. Our Magento abandonment guide walks through that order of work, and our look at one-step checkout extensions covers when buying one is genuinely the right call.
Not ready for that yet? Our free automated site audit scores everything that happens before checkout (homepage clarity, trust signals, SEO, performance and security), estimates the revenue impact of each finding, and walks a guest checkout as far as it can go without placing an order. What it cannot do is test the real flow behind your login, on real devices, against your own analytics. That is the boundary between the two, and the report says so.
Sources
Every figure above is drawn from one of the following primary sources.
- Baymard Institute: One Page Checkouts: the Holy Grail of Checkout Usability?
- Baymard Institute: Always Collapse Completed Accordion Checkout Steps into Summaries
- Baymard Institute: 50 Cart Abandonment Rate Statistics
- Shopify: Stellar Eats case study
- Digismoothie: One-page vs. multi-page checkout: real data analysis
- Shopify Community: merchant discussion of the one-page checkout default
Every statistic in this article is attributed to the primary source listed above. We have not added figures we could not trace, and we have left out one widely repeated claim for exactly that reason. Last reviewed October 2026.
